Multifactorial contributions to WINROP to enhance prediction of severe retinopathy of prematurity

Abstract

To the Editor, In the past decade, multiple algorithmic models to predict various diseases of pathological severity have been the focus of the medical and scientific community. In the present context, WINROP is an incomplete model to identify preterm infants at high-risk to develop threshold retinopathy of prematurity (ROP) and needs to be improved from its current predictive parameters (1). ROP is a multifactorial disorder involving numerous pathophysiological processes including high inflammatory state, oxidative stress, disordered metabolic state and impaired angiogenesis. We agree with Verd et al. (2) that breastfeeding is yet known to protect against ROP, and hence is a solid modifiable risk factor; accordingly, breastfeeding as well as early aggressive parenteral nutrition (3) are worthy of consideration in predicting ROP. This point has not been taken into account in the previous study (1) because all infants received mixed feeding at one time of their hospitalisation due to logistic organisation in this neonatal unit. Potentially, it would be relevant to integrate the type of nutrition and nutritional intakes comparing to the targets of the nutrition guidelines for preterm infants, realizing that weight gain and Insulin-like growth factor-1 do not fully reflect the benefits provided by breastfeeding, since immunologic advantages also appear to be fulfilled. In this context, the level and duration of supplemental oxygen in ROP must be factored in (1,3), as oxygen concentration is an established prominent predisposing factor, corroborated in numerous experimental animal models; hence despite limitations to which one can clinically limit oxygen supplementation, this factor cannot be undermined based on numerous meta-analyses (3). Adaptation to postnatal life, as reflected in the five minutes Apgar score may also be of interest (1), although the reproducibility of its importance may be overshadowed by other determinants in the clinical setting. Other factors that contribute to genesis of ROP such as inflammation (1,3) and hyperglycaemia (1) also seem relevant to introduce in a new model of prediction of severe ROP; yet, these factors require additional corroboration. Altogether, we agree with Verd et al. (2) that feeding type should be included in algorithms to predict ROP, but independently, would not be sufficient. Refined models would also enable to predict severe ROP, defined by threshold and pre-threshold ROP, as factors involved in pathologic progression differ from those simply implicated in relatively mild arrest in vascular development. Hence, multifactorial contribution needs to be accounted for over the sequential series of clinical events that predispose to ROP; this requires a more detailed evaluation of these criteria. Ideally, a large multicentric prospective study which would limit bias is needed to identify critical determinants for more accurate prediction of ROP.